[SOLVED] Rename a Cluster (Not a Node)

SpoonSauce

New Member
May 2, 2017
2
2
3
42
Is it possible to rename a cluster?

I've only tried a few basic things, including stopping all relevant services and modifying /etc/corosync/corosync.conf (cluster_name: MyNewCluster), but it seems to be pulling the config from SQLite once I start everything back up?!

Ideally I wouldn't have to re-create everything!

Thanks!
 
  • Like
Reactions: aasami
If you DO change the
```
totem {
cluster_name: <value>
```
clustername in /etc/pve/corosync.conf (as I did), DO restart *all* the corosync daemons on all the nodes (systemctl restart corosync) before (or after) the next cluster node restart. As the name is used to create the multicast address, this is also NOT a "run time" configuration change (will notice by chance it in the corosync/CMAP syslog messages), and will keep using the old name's multicast address, until you've restarted the crosync daemon, as those that started with the new name, and those still "running" since the old name, won't be able to communicate.

Hope this saves somebody some sweat and stress while having to find out during a production reboot why this rebooted node can't communicate with the rest with no errors :D
 
Thanks so much for that this really solved my problem of cluster join info. I found I had rename the cluster by accident on main node in the cluster. Fix that restart corosync on all nodes. Now it is fixed.
 
Doesn't the version have to be changed as well? I believe corosyc expects this which similar to making DNS changes.

totem {
cluster_name: clu3
config_version: 3 <----- This needs to be changed?
 
This works in 2024. Any other place where cluster-name is in use and should be replaced or any side-effects?
After changing in corosync-file i restarted corosync on all nodes -> and yes config_version should always be set to value +1
 
As described above, I changed the cluster name and configuration version in the /etc/pve/corosync.conf file. Then I restarted the service.
However, after that, in the ui, in the Datacenter (...) tree, I see the old cluster name. In order for the name to be updated, you need to restart the pveproxy service:
Bash:
systemctl restart pveproxy